Presentation is loading. Please wait.

Presentation is loading. Please wait.

15-441 Computer Networking Lecture 9 – IP Packets, Routers.

Similar presentations


Presentation on theme: "15-441 Computer Networking Lecture 9 – IP Packets, Routers."— Presentation transcript:

1 Computer Networking Lecture 9 – IP Packets, Routers

2 Lecture 9: Overview Last lecture How does choice of address impact network architecture and scalability? What do IP address look like? How to get an IP address? This lecture What do IP packets look like? How to handle some differences between LANs? How do routers work?

3 Lecture 9: Outline IP Packet Format Router Internals Route Lookup

4 Lecture 9: IP Service Model Low-level communication model provided by Internet Datagram Each packet self-contained All information needed to get to destination No advance setup or connection maintenance Analogous to letter or telegram versionHLenTOSLength IdentifierFlagOffset TTLProtocolChecksum Source Address Destination Address Options (if any) Data Header IPv4 Packet Format

5 Lecture 9: IPv4 Header Fields Version: IP Version 4 for IPv4 HLen: Header Length 32-bit words (typically 5) TOS: Type of Service Priority information Length: Packet Length Bytes (including header) Header format can change with versions First byte identifies version Length field limits packets to 65,535 bytes In practice, break into much smaller packets for network performance considerations ver- sio n HLe n TOSLength Identifier Fl ag s Offset TTLProtocolChecksum Source Address Destination Address Options (if any) Data

6 Lecture 9: IPv4 Header Fields Identifier, flags, fragment offset  used primarily for fragmentation Time to live Must be decremented at each router Packets with TTL=0 are thrown away Ensure packets exit the network Protocol Demultiplexing to higher layer protocols TCP = 6, ICMP = 1, UDP = 17… Header checksum Ensures some degree of header integrity Relatively weak – 16 bit Options E.g. Source routing, record route, etc. Performance issues Poorly supported ver- sio n HLe n TOSLength Identifier Fl ag s Offset TTLProtocolChecksum Source Address Destination Address Options (if any) Data

7 Lecture 9: IPv4 Header Fields Source Address 32-bit IP address of sender Destination Address 32-bit IP address of destination Like the addresses on an envelope Globally unique identification of sender & receiver ver- sion HLen TOSLength Identifier Fla gs Offset TTLProtocolChecksum Source Address Destination Address Options (if any) Data

8 Lecture 9: IP Delivery Model Best effort service Network will do its best to get packet to destination Does NOT guarantee: Any maximum latency or even ultimate success Sender will be informed if packet doesn’t make it Packets will arrive in same order sent Just one copy of packet will arrive Implications Scales very well Higher level protocols must make up for shortcomings Reliably delivering ordered sequence of bytes  TCP Some services not feasible Latency or bandwidth guarantees

9 Lecture 9: IP Fragmentation Every Network has Own Maximum Transmission Unit (MTU) Largest IP datagram it can carry within its own packet frame E.g., Ethernet is 1500 bytes Don’t know MTUs of all intermediate networks in advance IP Solution When hit network with small MTU, fragment packets Might get further fragmentation as proceed farther host router MTU = 4000 MTU = 1500 MTU = 2000

10 Lecture 9: Reassembly Where to do reassembly? End nodes or at routers? End nodes Avoids unnecessary work where large packets are fragmented multiple times If any fragment missing, delete entire packet Dangerous to do at intermediate nodes How much buffer space required at routers? What if routes in network change? Multiple paths through network All fragments only required to go through destination

11 Lecture 9: Fragmentation Related Fields Length Length of IP fragment Identification To match up with other fragments Flags Don’t fragment flag More fragments flag Fragment offset Where this fragment lies in entire IP datagram Measured in 8 octet units (13 bit field)

12 Lecture 9: IP Fragmentation Example #1 host router MTU = 4000 IP Header IP Data Length = 3820, M=0

13 Lecture 9: IP Fragmentation Example #2 router MTU = 2000 IP Header IP Data Length = 3820, M= bytes IP Header IP Data Length = 2000, M=1, Offset = bytes IP Data IP Header Length = 1840, M=0, Offset = bytes

14 Lecture 9: IP Fragmentation Example #3 IP Header IP Data Length = 2000, M=1, Offset = bytes IP Data IP Header Length = 1840, M=0, Offset = bytes host router MTU = 1500 IP Header IP Data Length = 1500, M=1, Offset = bytes IP Header IP Data Length = 520, M=1, Offset = bytes IP Header IP Data Length = 1500, M=1, Offset = bytes IP Header IP Data Length = 360, M=0, Offset = bytes

15 Lecture 9: IP Reassembly Fragments might arrive out-of- order Don’t know how much memory required until receive final fragment Some fragments may be duplicated Keep only one copy Some fragments may never arrive After a while, give up entire process IP Header IP Data Length = 1500, M=1, Offset = 0 IP Header IP Data Length = 520, M=1, Offset = 1480 IP Header IP Data Length = 1500, M=1, Offset = 1980 IP Header IP Data Length = 360, M=0, Offset = 3460 IP Data IP Data IP Data IP Data

16 Lecture 9: Fragmentation and Reassembly Concepts Demonstrates many Internet concepts Decentralized Every network can choose MTU Connectionless Each (fragment of) packet contains full routing information Fragments can proceed independently and along different routes Best effort Fail by dropping packet Destination can give up on reassembly No need to signal sender that failure occurred Complex endpoints and simple routers Reassembly at endpoints

17 Lecture 9: Fragmentation is Harmful Uses resources poorly Forwarding costs per packet Best if we can send large chunks of data Worst case: packet just bigger than MTU Poor end-to-end performance Loss of a fragment Path MTU discovery protocol  determines minimum MTU along route Uses ICMP error messages Common theme in system design Assure correctness by implementing complete protocol Optimize common cases to avoid full complexity

18 Lecture 9: Internet Control Message Protocol (ICMP) Short messages used to send error & other control information Examples Ping request / response Can use to check whether remote host reachable Destination unreachable Indicates how packet got & why couldn’t go further Flow control Slow down packet delivery rate Redirect Suggest alternate routing path for future messages Router solicitation / advertisement Helps newly connected host discover local router Timeout Packet exceeded maximum hop limit

19 Lecture 9: IP MTU Discovery with ICMP Typically send series of packets from one host to another Typically, all will follow same route Routes remain stable for minutes at a time Makes sense to determine path MTU before sending real packets Operation Send max-sized packet with “do not fragment” flag set If encounters problem, ICMP message will be returned “Destination unreachable: Fragmentation needed” Usually indicates MTU encountered host router MTU = 4000 MTU = 1500 MTU = 2000

20 Lecture 9: MTU = 4000 IP MTU Discovery with ICMP host router MTU = 1500 MTU = 2000 IP Packet Length = 4000, Don’t Fragment router ICMP Frag. Needed MTU = 2000

21 Lecture 9: MTU = 4000 IP MTU Discovery with ICMP host MTU = 1500 MTU = 2000 IP Packet Length = 2000, Don’t Fragment router ICMP Frag. Needed MTU = 1500 router

22 Lecture 9: MTU = 4000 IP MTU Discovery with ICMP When successful, no reply at IP level “No news is good news” Higher level protocol might have some form of acknowledgement host MTU = 1500 MTU = 2000 IP Packet Length = 1500, Don’t Fragment router

23 Lecture 9: Outline IP Packet Format Router Internals Route Lookup

24 Lecture 9: L L Router Architecture Overview Two key router functions: Run routing algorithms/protocol (RIP, OSPF, BGP) Switching datagrams from incoming to outgoing link input port output port Line Card

25 Lecture 9: Router Physical Layout Juniper T series Cisco Crossbar Linecards Switch

26 Lecture 9: Line Card: Input Port Decentralized switching: Process common case (fast-path) packets Decrement TTL, update checksum, forward packet Given datagram dest., lookup output port using routing table in input port memory Queue needed if datagrams arrive faster than forwarding rate into switch fabric Physical layer: bit-level reception Data link layer: e.g., Ethernet

27 Lecture 9: Line Card: Output Port Queuing required when datagrams arrive from fabric faster than the line transmission rate

28 Lecture 9: Buffering Suppose we have N inputs and M outputs Multiple packets for same output  output contention Switching fabric may force different inputs to wait  Switch contention Solution – buffer packets when/where needed What happens when these buffers fill up? Packets are THROWN AWAY!! This is where packet loss comes from

29 Lecture 9: Network Processor Runs routing protocol and downloads forwarding table to forwarding engines Performs “slow” path processing ICMP error messages IP option processing Fragmentation Packets destined to router

30 Lecture 9: Three Types of Switching Fabrics

31 Lecture 9: Switching Via a Memory First generation routers  looked like PCs Packet copied by system’s (single) CPU Speed limited by memory bandwidth (2 bus crossings per datagram) Input Port Output Port Memory System Bus Modern routers Input port processor performs lookup, copy into memory Cisco Catalyst 8500

32 Lecture 9: Switching Via a Bus Datagram from input port memory to output port memory via a shared bus Bus contention: switching speed limited by bus bandwidth 1 Gbps bus, Cisco 1900: sufficient speed for access and enterprise routers (not regional or backbone)

33 Lecture 9: Switching Via an Interconnection Network Overcome bus bandwidth limitations Crossbar provides full NxN interconnect Expensive Banyan networks & other interconnection nets initially developed to connect processors in multiprocessor Typically less capable than complete crossbar Cisco 12000: switches Gbps through the interconnection network

34 Lecture 9: Outline IP Packet Format Router Internals Route Lookup

35 Lecture 9: How To Do Longest Prefix Match 128.2/ / / /24 default 0/0 0 Traditional method – Patricia Tree Arrange route entries into a series of bit tests Worst case = 32 bit tests Problem: memory speed is a bottleneck Bit to test – 0 = left child,1 = right child

36 Lecture 9: Speeding up Prefix Match - Alternatives Content addressable memory (CAM) Hardware based route lookup Input = tag, output = value associated with tag Requires exact match with tag Multiple cycles (1 per prefix searched) with single CAM Multiple CAMs (1 per prefix) searched in parallel Ternary CAM 0,1,don’t care values in tag match Priority (I.e. longest prefix) by order of entries in CAM

37 Lecture 9: Speeding up Prefix Match - Alternatives Route caches Packet trains  group of packets belonging to same flow Temporal locality Many packets to same destination Other algorithms Routing with a Clue [Bremler-Barr – Sigcomm 99] Clue = prefix length matched at previous hop Why is this useful?

38 Lecture 9: Important Concepts Base-level protocol (IP) provides minimal service level Allows highly decentralized implementation Each step involves determining next hop Most of the work at the endpoints ICMP provides low-level error reporting IP routers Architecture Optimized for common case processing Complex/expensive lookup algorithms (especially in comparison to ATM fixed length lookup)

39 Lecture 9: Important Concepts IP forwarding  global addressing, alternatives, lookup tables IP addressing  hierarchical, CIDR, IP service  best effort, simplicity of routers IP packets  header fields, fragmentation, ICMP

40 Lecture 9: Next Lecture How do forwarding tables get built? Routing protocols Distance vector routing Link state routing

41 EXTRA SLIDES The rest of the slides are FYI

42 Lecture 9: IPv4 Header – RFC791 (1981) ver length 32 bits data (variable length, typically a TCP or UDP segment) 16-bit identifier Header checksum time to live 32 bit source IP address header length type of service flags fragment offset Protocol 32 bit destination IP address Options (if any) Padding (if any)

43 Lecture 9: ICMP: Internet Control Message Protocol Used by hosts, routers, gateways to communication network-level information Error reporting: unreachable host, network, port, protocol Echo request/reply (used by ping) Network-layer “above” IP: ICMP msgs carried in IP datagrams ICMP message: type, code plus first 8 bytes of IP datagram causing error Type Code description 0 0 echo reply (ping) 3 0 dest. network unreachable 3 1 dest host unreachable 3 2 dest protocol unreachable 3 3 dest port unreachable 3 6 dest network unknown 3 7 dest host unknown 4 0 source quench (congestion control - not used) 8 0 echo request (ping) 9 0 route advertisement 10 0 router discovery 11 0 TTL expired 12 0 bad IP header

44 Lecture 9: Line Cards Often uses special purpose hardware (e.g. ASICs) Network interface cards Fast path (common-case) processing Decrement TTL Recompute checksum Forward to next hop line card Forwarding engine

45 Lecture 9: Switch Buffering 3 types of switch buffering Input buffering Fabric slower than input ports combined  queuing may occur at input queues Can avoid any input queuing by making switch speed = N x link speed Output buffering Buffering when arrival rate via switch exceeds output line speed Internal buffering Can have buffering inside switch fabric to deal with limitations of fabric

46 Lecture 9: Input Port Queuing Which inputs are processed each slot – schedule? Head-of-the-Line (HOL) blocking: datagram at front of queue prevents others in queue from moving forward

47 Lecture 9: Output Port Queuing Scheduling discipline chooses among queued datagrams for transmission Can be simple (e.g., first-come first-serve) or more clever (e.g., weighted round robin)

48 Lecture 9: Virtual Output Queuing Maintain per output buffer at input Solves head of line blocking problem Each of MxN input buffer places bid for output Challenge: map bids to schedule of interconnect transfers

49 Lecture 9: Speeding up Prefix Match Cut prefix tree at 16/24/32 bit depth Fill in prefix tree entries by creating extra entries Entries contain output interface for route Add special value to indicate that there are deeper tree entries Only keep 24/32 bit cuts as needed Example cut prefix tree at 16 bit depth Only 64K entries

50 Lecture 9: Prefix Tree X73333XX Port 1Port 5Port 7 Port 3 Port 9 Port 5

51 Lecture 9: Prefix Tree Subtree 1Subtree 2 Subtree X73333XX95

52 Lecture 9: Speeding up Prefix Match Scaling issues How would it handle IPv6 Other possibilities Why were the cuts done at 16/24/32 bits?


Download ppt "15-441 Computer Networking Lecture 9 – IP Packets, Routers."

Similar presentations


Ads by Google